JP2008512672A - Apparatus and method for analysis of size, shape, angularity of mineral and rock pieces, and composition analysis - Google Patents

Abstract

Instruments and methods for three-dimensional measurement of size and shape, and for compositional analysis of mineral and rock fragments and similar objects. A mixture of small pieces or objects of the same or different sizes, such as minerals and rocks, are fed individually and automatically onto a conveyor belt, and are subjected to 3D mechanical vision measurements using a laser and two cameras, followed by visible Spectroscopic measurements using light and infrared light are made and then collected at the end of the conveyor. Computer software can be used to automatically perform measurements according to the built-in measurement process or user-specific methods, and to determine the size, shape, roundness, and preferably rock composition and other characteristics of each individual object. Calculate the statistical distribution of the relevant characteristics.

Description

  The present invention relates generally to instruments and methods for the analysis of mineral and rock fragments, and in particular to determine the size and shape of minerals and rock fragments, including the shape angularity (the shape / contour of the corners). And automatically analyze such pieces based on optical and spectrophotometric methods to determine their composition and classify the pieces based on size, shape and petrological type The present invention relates to a new apparatus and method.

  Traditional manual and mechanical methods of rock and mineral fragments (particles) size analysis, petrological descriptions, and other analyzes are well established, and the construction aggregate industry has For example, it is routinely used for evaluation of road gravel and concrete aggregate. However, conventional methods often require well-trained professionals and are nevertheless difficult to obtain reproducible results, which is laborious, time consuming and costly.

  Methods and instruments for optical analysis of mineral and rock fragments in the prior art are described, for example, in WO 02 / 090942-A1, DE 29800809-U1, DE 20117494-U1. No. 6,219,141-U1, U.S. Pat. No. 6,061,130, U.S. Pat. No. 5,309,215. However, the methods or instruments disclosed in the literature do not provide a three-dimensional analysis for the automatic sizing / shape determination of minerals and rock fragments (eg, classification by shape and angularity) nor analysis of rock composition .

  In order to determine the size, shape, and preferably composition, of mineral and rock pieces, there is a need for methods and instruments that can provide reliable and reproducible high-throughput measurements of such pieces.

SUMMARY OF THE INVENTION One object of the present invention is to provide mineral, rock, gravel, natural or manufactured or recycled to automatically determine the size and shape of each piece / object of a sample containing multiple pieces / objects. To provide an apparatus and method for high-throughput analysis of objects / pieces in a sample selected from selected aggregates and the like. Determining this shape includes parameters that define shape and angularity.

  The apparatus of the present invention comprises a suitable supply system for supplying said objects in a single layer flow of a certain maximum width such that the objects are spaced apart from each other (i.e. do not overlap or directly contact). A combination of an optical image detection system configured to obtain three-dimensional surface data for each object and a control system comprising processing means and dedicated software, the processing means and dedicated software of the control system comprising Process the image of the object, automatically determine the size and shape of the object, and return parameters representing the size and shape of the object. Preferably, the apparatus further comprises a spectroscopic detection system for obtaining a spectrum of the sample, wherein the spectrum of the sample is obtained by a control system to obtain data representative of the composition and / or petrological type of the sample object. It is processed. The apparatus and method of the present invention is capable of automatically and rapidly analyzing petrological samples, for example, to assess quality and suitability for construction, concrete, road construction, and related applications. . Such analysis has so far been based on laborious semi-mechanical methods (sieving and grinding) and manual visual inspection.

  Another aspect of the invention provides a method for determining the size and shape of an object in a plurality of objects in a sample selected from minerals, rocks, gravel, natural or manufactured or recycled aggregates, and the like. Providing and determining the shape, at least determining a shape parameter or class indicating whether the piece is long and / or flat, and determining angularity (angular contour) Including.

  Another aspect provides a computer program product for performing data analysis and determination in the methods described herein and for controlling data acquisition.

  The apparatus of the present invention can be configured for a particular sample type, but in general, for batch samples containing multiple objects / pieces such as minerals, rocks, gravel, natural or manufactured or recycled aggregates. Suitable for automatic analysis.

  As used herein, the term “shape” generally refers to the macroscopic shape and angularity of an object, and preferably also refers to surface texture. In this specification, the parameter “form” is used to classify the analyzed object according to the ratio of the main axis (long axis, intermediate axis, short axis) of the object. Accordingly, objects can be classified into four classes: long, flat and long, flat, and cube-like (cubic), as shown in FIG. 9c. If the length for all axes is substantially similar, the object is classified as cubic, the middle and minor axes are similar, and if the major axis is substantially longer, the object is longer If the major and intermediate axes are similar and are substantially longer than the minor axis, then the object is classified as flat and all axes are different, i.e. the major axis is more substantial than the middle axis. If it is long and the intermediate axis is substantially longer than the short axis, the object is classified as long and flat. Various cut-off values can be selected to define the boundaries between shape classes, and in one classification scheme, the two axes have a ratio of less than 0.67 (short axis: However, this value can be set to 0.5 or 0.75, for example.

  These various shapes can also be represented by flakeness ratios or elongation ratios. Generally in the art, depending on which standard is used, a piece is considered flaky if its thickness is less than 0.5 or 0.67 of its width. If the width is less than 0.5 or 0.67 of the length, it is considered “long”. Such classification is easily achieved using the method of the present invention, where measured objects are grouped according to the measured principal axis ratio and conform to relative measurement classification using conventional caliper measurements. This is accomplished by selecting an appropriate interval so as to define a group to perform.

  The apparatus comprises a feeder (feeding device) for feeding the sample objects onto the conveyor belt so that they are separated from each other in a single layer flow and do not overlap each other or do not directly contact each other. In some embodiments, the objects are substantially aligned to one line so that the sample objects enter the detection area of the device one by one, but the detection means and analysis software distinguish between the objects. It is also possible to detect two or more objects at the same time as long as it can be performed. Several feeder means known to those skilled in the art can be used in accordance with the present invention, such as one or more conveyor belts arranged at an angle, eg, perpendicular to the main transport belt of the apparatus. Can be used. The presently preferred embodiment comprises an oscillating spiral elevator 8 as shown in FIG. 2, the dimensions of which depend on the particular sample being analyzed. A suitable spiral elevator is provided by ScanVibro, Denmark. Other configurations may be used, such as, for example, a vibrating dosign feeder provided by Vibratechniques Ltd (Vibratechnics) (UK).

  The main conveyor 11 used in the apparatus can be easily selected by a person skilled in the art, preferably using a suitable high-precision motor to ensure that the conveyor speed is fixed during operation of the apparatus. Should be driven. In one embodiment, an electric motor is used and its speed is directly dependent on the input AC frequency, which is accurately controlled using an adjustable frequency controller. In the configuration shown in FIG. 2, the supply conveyor 12 feeds objects from the feeder 8 to the main conveyor 11 and the supply conveyor is set to low speed to help distribute the objects discretely. The main conveyor 11 is driven by a roller 10 whose shaft is connected to an electric motor fixed to the distal panel 14 of the cover 13. The open side of the cover 13 can also be closed using an attachable front panel (not shown).

Optical detection and surface analysis The apparatus comprises an illumination source that projects a planar beam of light (ie, collimated, preferably coherent light) across the main transport belt, typically the beam of motion. It is also perpendicular to the direction and preferably to the plane of the belt. In one embodiment, coherent diode laser light is used with a suitable lens to produce a collimated planar coherent light beam. The width of the beam across the belt determines the maximum width of the object to be analyzed, typically a beam width of about 40-160 mm is appropriate for conventional rock aggregate samples, for example about 40- For example, a range of 100 mm, a range of about 40 to 80 mm such as about 50 mm or about 60 mm, or a wider range of about 100 mm, 120 mm, 140 mm, or 160 mm. If the long axis of the object is along the belt so that the width of the object across the belt is less than or equal to the beam width, longer objects can be analyzed. The width of the beam across the beam plane itself affects the resolution of the surface analysis, and preferably the width at the maximum surface height of the object to be analyzed is approximately 0. 0 mm, for example 0.05 mm or less. It is less than 1 mm. However, it is difficult to define the beam width. That is, the observed width depends on the ambient light level, the texture of the surface on which the light is projected, etc., and in particular the lens configuration, i.e. the beam is, for example, 30-40 mm from the conveyor belt. Or, for smaller pieces, it will depend on a lens configuration that is appropriately configured to be focused at a lower estimated piece height.

  The present invention is particularly useful for the analysis of gravel-sized aggregates (about 2-4 mm to 40-60 mm) and pebbles-sized aggregates (about 20-40 to about 160 mm), and sand-sized pieces ( Such suitable solutions are well known to those skilled in the art, although different feeder configurations may be required.

  The image capture means comprises a machine vision camera having at least one image detector, typically a lens and a CCD or CMOS detector, the lens having a suitable focal length The one is selected. The detector (s) are arranged to capture a diffuse reflection image of a planar light beam that illuminates an object conveyed through the beam path. Image detectors can capture images at short regular intervals, such as about 10 images per second, or more, for example about at least 20 images per second, including at least 24 or 26 images per second. Is controlled to capture. For faster processors, use a fast rate that can get about 25-120 images per second, such as about 40 or 48 images per second, or up to 75 or 100 images per second. Can do. The image capture period determines the allowable speed of the conveyor and hence the measurement throughput. For high throughput analysis, a high image capture period, such as at least 50 images per second, at least 100 or 120 images per second, is desirable. The software and control system of the present invention enables high-throughput analysis and, using current system configurations, at least about 100 objects per hour, preferably up to about 100 objects with a maximum width diameter (major axis) of less than 100 mm. It can be configured to analyze at least 200 rocks with a large diameter less than 100 mm per hour. However, if a longer object is placed parallel to the conveyor belt and the width of the object relative to the lateral direction of the belt is less than the width of the beam, such a long object can be analyzed. Accordingly, the feeder of the present apparatus is preferably configured to place the object on the conveyor with its longest axis along the direction of belt movement.

  In a preferred embodiment, at least two image detectors are used, preferably the image detector is configured to capture an image of the reflection of the planar beam, and the detector is relative to the object from different angles relative to the planar beam. Preferably directed to the object, one on each side of the planar beam, with each of the image planes having a horizontal axis perpendicular to the direction of the conveyor belt and parallel to the planar beam Arranged to have. In other words, the detector is preferably aligned with the direction of the belt, one on the front of the beam and the other on the opposite side of the beam, as shown in FIGS.

  The apparatus further comprises a control system comprising processing means in which a computer program for controlling the mechanical part and the hardware part of the device is stored, and a memory for storing the captured image. The system is adapted to process a series of images of the object to automatically determine the size and shape of the object, including the size and shape of the object, and preferably the sample, including angularity and shape Returns a parameter representing the petrological composition of the object.

  In one embodiment of the apparatus, the control system is adapted to process a series of images of the object and the resulting three-dimensional surface image based on a computer program, one of the following parameters: Or calculate multiple things. Object size: major axis, intermediate axis, minor axis length: elongation ratio and flake ratio: shape class: equivalent shape index: equivalent flake index: sphericity, roundness or angularity Value and / or class: Statistical distribution of one or more of the parameters for a plurality of analyzed pieces.

  In a preferred embodiment, the control system is adapted to perform a mathematical morphology algorithm to process a series of images and is correlated with roundness or angularity parameters and / or classes pseudo volume loss Calculate the value.

  Rock piece angularity is traditionally described using one of six categories on the Powers scale, originally proposed by Powers, 1 is “very square”, 2 is “square”, 3 is “somewhat square”, 4 is “somewhat round”, 5 is “round”, and 6 is “very round” (Powers, MC "A new rounds scale for secondary particles" Journal of Secondary Petrology (1953), Vol. 23, pages 117-119).

  As will be described in more detail herein, the angularity scale can be interpreted as being based on the angularity measurement according to the present invention. This scale may be based on the determination of “% volume of angles”, in which the object is substantially as if classified by manual inspection according to the Powers scale. Appropriate cut-off values can be selected to produce an interval for classification. We use the apparatus and method of the present invention to automatically classify pieces into classes according to Powers' angular definition, using good statistical correlation to independent classification by manual inspection. Found that it is possible.

Morphology Analysis The two-camera laser triangulation system of the preferred embodiment of the present invention enables sufficient surface coverage and effectively provides a complete model of the upper hemisphere of a piece. Thus, each piece can be represented as a range image without the effects of occlusion and perspective, as shown in FIG.

There are natural similarities between range images and grayscale morphology. Is collected using laser triangulation, cloud data represented by the distance image is a free surface in fact, K ³ space. In order to characterize the morphology of the pieces, we need to assume that the object is an individual in an obstructed area below the visible surface. Gray-level image represented in K ² space function f (x, y) as is the gray scale morphology, a set of points in ^{K 3} spaces [x, y, f (x , y)] to be considered. The main shadow portion U [f] can be defined as follows.

U [f] = {p (x, y, z): z ≦ f (x, y)} (Formula 1)

Therefore, the main shadow portion of the set X in the K ³ space is the volume of the points included in the shadow of X. In this case, the light source is a point light source at an infinite distance in the positive z direction (Standley R., et al. Sternberg, SR Grayscale Morphology CVGIP 35 (1986), pages 333-355). This corresponds to the direction of illumination in the laser triangulation system, so the shadow is the same as the shadow cast by the laser.

  Then gray scale erosion and dilation are defined as follows: Erosion of the image f using a structuring element g involves moving the structuring element to each point of the image and taking the “minimum difference”. Therefore, it becomes as follows.

  Similarly, for grayscale expansion, we move the structuring element to each point on the image and take the “maximum of sum”. Therefore, it becomes as follows.

These operations, in their standard form, are extremely sensitive to noise and local intensity variations. This is a particular problem when working with distance images. This is because some kind of impulse noise or point occlusion due to laser speckles may greatly affect the processing result. In soft morphology, the maximum and minimum operators used in dilation and erosion are replaced with more general weighted order statistics. The structuring element consists of a “hard” center and a “soft” boundary, which is weighted to affect the operation more than the boundary. Therefore, the inventors define a structured system [B, A, k] consisting of a finite set A and B, A⊂B, and three parameters of a natural number k satisfying 1 ≦ k ≦ | B |. . B is a structured set, A is its hard center, and k is its degree index or iteration parameter. We represent the k-th minimum member and the k-th maximum member of the set as min ^(k) and max ^(k) , respectively, in accordance with the convention used by Koskinen et al. ( Koskinen et al., Soft Morphological Filters Proceedings of SPIE (1991), 1568, 262-270), we use <> to indicate iterative operations.

  The soft erosion of image f using the structured system [B, A, k] is then best defined by the following equation:

  Similarly, the soft extension of f by [B, A, k] is defined as follows:

  Careful selection of appropriate structuring elements is important for the effectiveness of morphological operations. In the analysis of an object according to the invention, the structuring element needs to be invariant to both the direction of the piece and the inclination of the piece surface at any point. Also, a morphological opening should result in minimal loss of volume in a piece with a sufficiently round surface. Under these constraints, a properly sized sphere is an ideal choice. Again, this is represented as a distance image. When using this approach, in practice, a discontinuous approximation to the hemisphere is used rather than a real continuous sphere. Applying a morphological closing or closing to a grayscale image using a structuring element with gray values representing the surface of the hemisphere is a rolling ball. Commonly known as a transform, this was first described by Sternberg (supra) as a means to smooth discontinuities in standard intensity images and to remove noise.

  This technique can be explained as follows. In the case of morphological closing, imagine rolling a ball over a three-dimensional surface described by the gray levels present in the image. If the surface is smooth with a relatively small curvature, the ball will come into contact with the surface at every point. However, the ball will not contact surface points in narrow holes or grooves, i.e. surface points contained where the curvature of the recess exceeds the curvature of the sphere. The resulting function is the union of the paths that the ball can follow, or more specifically the union of the transformations of the spherical structuring elements for every point on the surface. That is, any narrow holes or grooves in the surface are “closed”. Similarly, morphological opening can be visualized as rolling the ball under the three-dimensional surface. In this case, what is lost with this procedure is a high curvature protrusion such as a sharp edge or corner. This is illustrated in FIG. The original surface, shown as a cross section, can be seen in FIG. 5a. A transformation is then applied (FIG. 5b). Any points on the surface that are not accessible by the sphere are deleted (FIG. 5c).

  Based on this, we can use the appropriately sized spherical structuring element to open a small range image that can provide a means to distinguish between different angular classes. This distinction can be provided simply by analyzing the ratio of volume lost ("volume of angles") by this procedure. This effectively simulates the natural wear process of rounding rock pieces such as sand and gravel by reducing each piece to a “sufficiently round” state. We want to propose that in this case the boundary of the structuring element should have an equal influence on the center and therefore A = B with respect to equations (4) and (5). The rank parameter k can then be selected according to the expected level of noise in the range image.

  Obviously, the ability to discriminate by this approach will largely depend on careful selection of the diameter of the spherical structuring element used. In fact, the original method proposed for manual petrological classification defines angularity for the size of the pieces. Wadell (Wadell, H. “Volume, shape and roundness of quartz particles”, from which the Powers scale (Powers, supra), which is generally regarded as authoritative and widely used, is derived. The technique used by Journal of Geology (1935), 43, 250-280) involves dividing the average radius of the corners of a two-dimensional granular image by the radius of the largest inscribed circle. Given that typical aggregate samples can contain very different sized pieces, the size of the structuring element seems to be chosen dynamically according to the size of the piece being analyzed. . However, with further considerations, the inventors have questioned the use of spherical structured elements. Initially, the height of the piece considered as a dimension orthogonal to the image plane has no effect on absolute volume loss. Pieces having a proportionally higher height do not lose much volume proportionally. In addition to this, the angularity of a corner is conventionally measured with respect to the maximum inscribed circle in a given projection (Wadell, supra). However, a suitable ratio of spheres for one projection may be inappropriate for another projection. Therefore, to remove the bias associated with the stretch ratio and flatness ratio of the pieces, we replace the spherical structuring element with an ellipse having the same aspect ratio as that of the piece being analyzed. The axes of the structured element are always kept aligned with the axes of the piece during this procedure, ensuring that the proportionality from any two-dimensional projection is maintained.

  Since the size of the aggregate piece is seen as a fundamental characteristic and is determined as a matter of course, it is easy to adaptively generate a similar proportion of structured elements. By experimentation, the inventors have found that optimal results are obtained with a structured element having a half axis that is 1/6 of the length of the half axis of the piece. It is also contemplated that a structured element having the same aspect ratio as the piece being analyzed but having another proportionality constant can be used, for example, about 1/4, about 1/5, about 1/7, about Use a structured element with a semi-axis having a value in the range of about 1/3 to 1/10, such as a range of 1/4 to 1/10 of the axis of the piece, such as 1/8. It is also contemplated. FIG. 6 shows a cross section of three pieces derived from the captured range image. The outer line represents the original boundary and the inner line represents the boundary after the morphological opening. FIG. 6a is taken from a sufficiently round piece. It can be seen that the boundary remains substantially unchanged and the volume loss is negligible.

  Note that when looking at a cross-section such as that shown in FIG. 6, the volume loss does not necessarily correspond to a visible profile. For example, in FIG. 6c, the volume loss below the left side of the profile corresponds to a corner that is not visible in this projection.

Texture analysis The image acquired as described above can also be used to determine the finer texture (roughness) of the piece, and preferably also the intermediate texture (porosity). it can. This is easily accomplished by analyzing the profile across at least a portion of the edge of the piece in the image, or across the upper hemisphere of the piece, and the roughness (wavy) of the edge causes a fine texture (smooth / Coarse) will be shown.

Absorbance / Reflectivity Measurement and Classification A useful embodiment of the device detects a reflection of the beam from a light source that sends a beam of visible and / or infrared light and an illuminated spot of the object, and And / or a spectrophotometric detector for measuring an infrared absorbance or reflection spectrum, wherein the control system is adapted to process the spectrum and to convert the spectrum into a reference spectrum based on a reference material and / or Compared to the spectral values, it is adapted to classify said objects according to a predetermined class system of mineral and rock pieces and similar object types and / or variants. Thereby, the apparatus and method of the present invention can automatically classify rock samples, aggregates, etc. according to a predetermined classification system.

  The apparatus is preferably capable of classifying samples according to the classification system described in European standard EN932-3, so that at least 10 of the more general classes, preferably at least 15 and 20 of the classes mentioned above. A sample object representing an individual is automatically determined.

The classification system groups rock types as follows:
Plutonic igneous rock: granite, syenite, granodiorite, diorite, gabbro; semi-deep igneous rock: coarse grain basalt, diorite; eruptive igneous rock (volcanic rock): Rhyolite, rough rock, andesite, quartz andesite, basalt; sedimentary rocks (divided into two groups based on their origin): clastic rocks: sandstone, conglomerate, breccia, granite rock, gray wacke, quartzite Non-clastic rocks, chemical rocks and biogenic rocks: limestone, chalk, dolomite, char; metamorphic rocks: amphibolite, gneiss, granulite, hornfels, calcite / dolomitic marble marble), quartzite, serpentine, schist, slate, mylonite.

  While it is highly preferred that the apparatus and method of the present invention be able to classify any given sample object into any one of the above classes, in some practical applications, Each of which can contain one or more of the above classes, the included classes of samples having similar aggregate properties so that they are distinct for use in industrial / construction. It would be very useful to easily determine several main classes that do not need to be done.

  For the analysis of the spectrum and the comparison to the reference spectrum and / or values, two main approaches are preferred: an approach using a statistical method using a statistical classifier or an approach using a neural network.

Neural networks consist of interconnected networks of their basic units, so-called neurons. The neuron receives an information vector containing R different values and multiplies each value p _i by an individual weight w _i . The sum of these products, the dot product of the p and w vectors (possibly plus the bias b), is the input to the transfer function f of the neuron that returns the output value a. The network is “trained” before performing the actual sample analysis, and the product of this training can expect to return the correct value, ie the class number, given the input value for the unknown sample. A collection of adjusted weights that can be made.

  The inventors have found that it is useful to apply a statistical method using the wavelet method in the method of the present invention. Wavelets have suitable local properties and have been found suitable for statistical modeling of high-dimensional data. In one embodiment, Daubechies wavelets with two vanishing moments are used. Classification accuracy can be improved by using several measurements from various locations of each reference sample object. For such average measurements, a reliable prediction of class membership can be easily derived by using wavelets.

  The visible and infrared light sources preferably provide visible and infrared light comprising a wavelength range of about 340 to about 4000 nm, such as preferably about 340 to about 1200 nm, and suitable for detecting said range The correct detector is selected.

  Suitable spectrophotometric detectors can be readily selected by those skilled in the art, for example, Avantse (Aerbeek, The Netherlands), which provides a suitable light source that can be selected to provide light in a suitable wavelength range. From AvaSpec spectrophotometer.

  The novel aspects of the present invention relating to spectroscopic analysis can be implemented as a stand-alone embodiment, i.e., the necessary features described above for spectroscopic analysis of objects and automatic rock analysis as described herein. It should be noted that it can be implemented as a device comprising. Such an apparatus generally comprises a suitable feeder and conveyor, preferably as described for the above-described embodiments, a light source providing a beam of visible and / or infrared light, and as described above, A spectrophotometer detector for sequentially detecting the reflection of the beam from an illuminated spot of an object and measuring a visible and / or infrared absorbance or reflection spectrum, the apparatus comprising: As described in more detail in, with a control system for spectral analysis and comparison with a reference spectrum / value.

  The present invention, in another aspect described above, is described herein for automatic analysis of image data as described herein for automatic analysis of sample objects as defined above. A computer program product for performing data analysis in the method and preferably for controlling data acquisition is provided. The computer program product includes instruction means for instructing a computer processor to do the following when loaded and executed on a computer.

Input data from at least two image detectors configured to capture a sequential image of reflections of a planar coherent beam that illuminates across the direction of flow of an object moving at a predetermined speed with a minimum predetermined frequency; Receive
-Storing the image of each object, processing the image to obtain data indicating the size of the object and three-dimensional surface data representing the object, or contour map and topographic data;
-Determining one or more size and shape parameters for the object based on the obtained data, wherein the shape parameter is whether the piece is long and / or flat; Shape parameters and / or classifications, and angularity / roundness parameters and / or classifications.

The computer program product preferably uses any or all of the analysis methods and morphological algorithms described above.
In some embodiments, the computer program product can be separated into substantially separate units. That is, one unit for data acquisition and measurement device control (conveyor speed, calibration of detectors, etc.) and another unit that can run on another computer for data analysis and data presentation, Can be separated.

  In useful embodiments, the computer program product further includes visual and / or obtained for each object as described above to determine petrological or other compositional information as described above for the object. The IR spectral data is input and adapted to be analyzed according to the methods described herein. Preferably, the information includes a classification of the object according to a predetermined class system of petrological rock types and / or variants, as described in more detail above.

  The invention is not limited to the specific disclosure herein and includes modifications within the scope of the appended claims. For example, having different light wavelengths, different lenses or reflective probes than those specified here for spectroscopic measurements, and having an additional camera to obtain additional information on surface texture properties and colors Having a system for sorting objects after measurement and according to measurement results, having different types of feeders and systems for controlling the entry of objects into and out of the conveyor, For this reason, it is possible to increase the number of aligned object rows to more than one. The instrument of the present invention can also be used to measure the size, shape, composition of other materials, such as ores, drilled core specimens or crushed, fish It can be used for the measurement of things like otoliths, recycled building materials, industrial minerals, decorative stones, chipping, metal minerals and alloys.

Examples Example 1: A guide program structure for a program of size, shape, angularity analysis part is shown diagrammatically in FIG. The following sections briefly describe the role and function of each section of the initial version of the software embodiment.

  Calibration of camera 1. A single frame is captured. For each column of the image, the brightest pixel (theoretically corresponding to the center of the laser line) is located. When the intensity of a pixel exceeds the “minimum intensity” (used to distinguish between a pixel that may be of the laser line and a pixel that corresponds to the background light level), the high at that point In order to calculate the length, the “triangulation algorithm subVI” is called. Thus, a 1D array of height values is generated for that frame. A median is taken to give the belt height. This is used as a reference value in all subsequent triangulation calculations.

  The median is taken because the majority of the columns in the image frame should be covered by the laser line. Therefore, selecting the median means that the height value corresponding to the blocked line is not selected. In addition, outliers caused by dust and the like are also excluded.

Calibration of camera 2. Same procedure as described above for camera 1.
"State machine"
State 0: Responds to a “stop” command. Close image, clear memory, etc.

  State 1 (initial state): Frames are continuously acquired from the camera 1 (parallel to the moving direction). In order to count the number of obstructed lines in each frame, the “occlusion-based piece detection means subVI” is called. If the number of occluded lines exceeds the “occlusion trigger level”, indicating that an approaching piece is blocking the laser line from the camera, the system moves to state 2. When stop is pressed, the system moves to state 0.

  State 2 (Acquire Data): Acquire frames continuously from both cameras. Call “calibrated triangulation subVI” to generate an array of height data. In order to merge the height data generated from each camera, “subVI to merge lines” is called. This routine uses a Boolean latch when the system starts to acquire data from the rock surface (ie, the height value returned by "calibrated triangulation subVI" is minimal). This is indicated when the height threshold is exceeded. This is initially set to false and is latched true immediately when a positive height value is returned.

  Stop criteria 1: If the latch remains false beyond a certain number of lines ("maximum inactive line" control), there must be clearly no piece under the laser and the trigger is false Absent. In this case, the routine stops acquiring data and returns to state 1.

  Stop Criteria 2: If the latch is triggered (i.e. set to true), but no camera has captured a positive height value, this should indicate the end of the piece. However, this may be caused by an occlusion middle piece, so some tolerance should be allowed. Thus, the counter is incremented when these conditions are met each time a pair of frames are captured. Thereafter, if either camera acquires positive height data, the counter is reset to zero. If the counter exceeds a predetermined value (“grace count”), the system stops acquiring and proceeds to state 3.

  Stop criteria 3: The system switches directly to state 0 whenever the stop button is pressed.

  Stop Criteria 4: Whenever an acquisition error occurs, the system stops acquiring and prompts the user.

  If fewer lines are acquired in this state than "Minimum Acquired Lines" (user control), the system will return to State 1 instead of State 3. This indicates that the piece is too small to process.

  State 3 (Process Range Image): This state takes as input the merged range image of the array format and takes all the subVIs (sub “virtual” needed to quantify the properties of the pieces. a subroutine called "instruments (virtual device)" (see Labview software environment from National Instruments)). There is also an option for the user to select and view any height profile across the piece, which can be useful when setting up the device to perform a diagnosis.

Guide to subVI Occlusion-based chip detection means subVI: This VI is used to detect the presence of rocks approaching the laser line. This works by monitoring a laser line image (in array format) seen by a camera facing the direction of motion. For each column of the image, the brightest pixel is located. If the intensity of the brightest pixel in a particular column is less than the “minimum intensity” (a threshold determined by the user), the column is said to be blocked. Thus, for each frame captured by the camera, this VI returns the number of occluded columns.

  Triangulation algorithm subVI: This software part implements an algorithm to calculate the spatial coordinates of one image point. Inputs are f (focal length of camera lens, unit is mm), d (distance between center of lens measured along camera axis and conveyor belt, unit is mm), theta (triangle angle ( triangulation angle (unit: degrees), j (pixel coordinates in the vertical direction, measured from the upper left of the image by default), i (pixel coordinates in the horizontal direction, measured from the upper left of the image by default), camera Sensor parameters (physical dimensions, units are mm and number of pixels). This VI returns the y and z coordinates of the point, the x coordinate being determined by the position of the conveyor belt.

  Calibrated triangulation subVI: This takes a single image frame (array format) and locates the center of the laser line in each column using the “centroid” technique. Next, this VI calls the “triangulation algorithm subVI” for each identified peak value and returns one line of height data.

  SubVI to merge lines: This takes the two lines of height data generated by “calibrated triangulation subVI” for each of the two cameras and merges them into a single line of height data To do. Before performing this VI, one of the lines must be reversed (the two cameras point in opposite directions) so that a particular element of each linear array corresponds to the same spatial position. , Must be aligned. The two lines are then merged point by point according to the following criteria:

Does the height value returned by each camera differ by more than the “maximum difference”?
If no: take the average of the two values.
If yes: is any value 0?
If yes: The point is blocked by one camera. The largest of the two values is taken as the height at that point.
If no: the difference is probably due to specular highlights or interreflections, which generally results in false high values. The smallest of the two values is taken as the height at that point.

  Any pixels having a height value of 10 or less are then set to 0 corresponding to the belt reference height level.

  Width check subVI: This is used to help align and calibrate the two cameras. The input is the height data of one line generated by “calibrated triangulation subVI”. This VI returns the width of the piece in terms of the number of pixels and the index of the farthest left position of the piece. This VI is used for the lines returned by both cameras to eliminate the difference.

  MM occlusion removal closing subVI: This VI uses gray level morphological closing to eliminate any discontinuities (holes) in the boundary of a piece caused by the point being blocked from both cameras. Is. The structured element is designed to have minimal impact on real surface features.

  MM spike removal opening subVI: This VI uses gray level morphological opening to eliminate any spikes (impulse noise) on the recovered surface due to mutual reflection from the concave surface It is. The structured element is designed to have minimal impact on real surface features.

  MinBoundingBox Auto subVI: This VI determines the length and width of the piece. The acquired distance image is a threshold (reduced to a binary representation, one value representing a small piece and the other representing a background). The orientation of the piece is determined using a standard IMAQ vision function centered on moment analysis. The piece image is then rotated to align its principal axis with the image coordinate system. The smallest enclosing rectangle is fitted to the image. The longer side of the rectangle is taken as the length of the piece and the shorter side is taken as the width.

  SubVI to calculate sieve size: This VI uses a calculation technique to calculate the minimum sieve size that a piece can pass from its calculated width and height. The volume of the piece as an equivalent ellipsoid is also calculated.

  SubVI for collating shape results: This VI takes numerical values for length, width and height and converts them into a format suitable for display.

  Image resampling subVI: This reduces the resolution of the image array by a specified reduction factor before determining the angularity of the piece.

  RollingBall controller subVI: This VI provides a framework that can undertake the operations necessary to calculate angularity. This VI calls "adaptive ellipse calculation means subVI" and generates an elliptical structured element according to the size of the piece to be analyzed. The “rolling ball algorithm subVI” is then called to perform the morphological opening. The volume loss of the piece is then converted to a Powers scale value.

  Adaptive ellipse calculation means subVI: Generates a semi-ellipsoidal structuring element whose principal axis is 1/6 of the principal axis of the piece (after re-sampling)

  Rolling ball algorithm subVI: Performs a morphological opening on a given image array using a given structured element.

  Sieve Size Distribution subVI: Updates the cumulative piece size distribution (represented as a separate array of 0.5 mm increments) each time a piece is processed.

Example 2: Angularity analysis of rock pieces A total of 200 rock pieces in the 8-32 mm size range were scanned using a laser triangulation system and analyzed by an elliptical morphology algorithm. Prior to the analysis, the pieces were visually assessed and graded manually according to the Powers scale by geologists from two independent institutions.

  FIG. 6 is a visual (manual) rated angular for each piece, expressed as a number between 1 (very angular) and 6 (fully rounded) according to the Powers scale. Shows a scatter plot of pseudo-volume loss due to morphological opening to the utility.

  The first observation is that the deviation in results is very large and appears to increase with angularity. This can be confirmed by plotting the standard deviation of the percent volume lost under each class against the Powers number, as shown in FIG. 7b. It can be seen that the standard deviation increases at a fairly uniform rate as the angularity increases, as also shown in FIG. 7a which shows the distribution of measured volume loss within each class. This indicates that the deviation is due to the difficulty of manually classifying angularity objectively, which means that a sufficiently round piece can be easily identified as such. It can be done, but it is much more difficult to distinguish between more square pieces. To examine the effectiveness of the algorithm of the present invention, the relationship between the average volume loss percentage and the Powers number was determined for each category. This is illustrated in FIG. The correlation is clear, and there is an approximately linear relationship between the average volume loss percentage and the manually evaluated Powers number. From this result, a correlation value of 0.987 is obtained.

Example 3: Rock analysis A method has been developed to obtain information about the texture of analyzed rock samples. Textures are often affected by the size of individual crystals, and therefore the rock crystal size can vary depending on their type, thus providing information that can be important for classification.

  In short, the method works as follows. That is, the method locates a sample in a two-dimensional image taken by a digital camera (eg, a CCD or CMOS type industrial camera) located above the object (eg, a laser beam and camera assembly is After being used for angularity evaluation), start by selecting a portion from the middle to avoid background and shadows. The photograph is then converted into a gray scale image and the number of gray tones reduced between 2 and 25. Experiments were conducted with various numbers of colors and this 2-25 interval that would give good results, with the best results appearing to be within 4-8 intervals.

  After reducing the color, each pixel is evaluated and a so-called “co-occurrence matrix” is created that contains information about the likelihood that two adjacent pixels are the same color. . If the rock has large crystals, it can be intuitively understood that adjacent pixels are likely to have the same color, and vice versa.

In the coincidence matrix P, each entry P _ij represents the number of cases (cases) in which a pixel with a grayscale value i is within a predetermined distance d from a pixel with a value j, where d is usually the pixel on the right , Or perhaps a collection of vectors, so that all of the surrounding pixels can be evaluated. P diagonal entries, ie all entries P _ij with i = j represent the number of cases where the compared pixels have the same color. Thus, for samples with large crystals, the maximum value in P is located diagonally or close to the diagonal, whereas for samples with small crystals, the values are more uniformly distributed over the matrix.

  After the co-occurrence matrix P was constructed, information about the texture was derived by evaluating the distribution of values in P roughly using the following equation:

  In addition to being photographed as described above, the samples analyzed in this example were further spectroscopically measured. The samples were of four different rock types, two of which were divided into two variants. The four rock types were basalt, gabbro, rhyolite and granite. Basalt and granite samples were further divided into two categories or variants. Thus, the samples represented 6 different rock samples, each containing 10 sub-samples (sub-samples). See Table 1. The sample is in two different spectral fields: a 400-1100 nm region or range, referred to herein as the visible range (VIS), and a 1000-3000 nm range, referred to herein as the near infrared range (NIR). Range, and finally the mid-infrared range (MIR) of 3000-30000 nm. Each sub-sample was measured at one spot in VIS and three spots in MIR and NIR. A total of 60 measurements were obtained with VIS and 180 with MIR and NIR, respectively.

Result Visible region (VIS)
The spectrum of the sample (sample) is fairly uniform with a fairly constant slope and several “valleys” and “hills” in the visible region (400-1100 nm herein), but its classification Was very accurate. With a limited number of measurements, ie 60, the accuracy of about 80 percent is probably higher than expected. It should also be easier to improve the classification by making more measurements.

  A common factor in classification in all of the spectra, namely VIS, NIR, and MIR, is that the most frequent error occurred when classifying samples of class or numbers 4 and 5. In those cases, class 4 samples were classified as belonging to class 5 with almost no exception, and vice versa. Again, in view of the fact that Class 4 and 5 samples are often considered as similar types of aggregates (of the same rock type), these two classes are considered as one. Classification was made. Thereby, the classification accuracy has increased dramatically, rising to 98 percent, but a more realistic accuracy will probably be between 90 and 95 percent.

Near infrared region (NIR)
The spectra in the near infrared region are very uniform, but have more obvious features that can be used to distinguish between classes by looking at their spectra. In short, a classification accuracy of 95 percent to 98 percent can be obtained very easily, and errors only occur in classes 4 and 5. The accuracy after reducing these to one class was 100 percent without exception.

Example 4
Automatic Analysis and Data Presentation User-friendly programs and computer interfaces have been developed to present data collected from high-throughput analysis of rock pieces using the apparatus shown in FIG. A screen view showing an example of output from a computer program is shown in FIG.

(I) Size determination The program determines the main axis of each object from the acquired image, the long axis (L) is the longest distance between two points of the object, and the intermediate axis (I) is relative to the main axis. The shortest axis (S) is orthogonal to the other two axes.
The size of each piece is indicated using the volume parameter, ie the calculated volume of an ellipse whose external dimensions correspond to the principal axis. This determination thus substantially corresponds to the size measurement defined in the European standard EN933-1.

(Ii) Shape classification Based on the ratio between axes, each object is in one of four groups: long, flat and long, flat, cubic. being classified.
If the ratio S / I (“flakiness ration”) and I / L (“elongation ration”) are both greater than 0.67, the object is cubic and I / L ≧ 0. If 67 and S / I <0.67, the object is flat; if I / L <0.67 and S / I ≧ 0.67, the object is long, I / L <0.67 and When S / I <0.67, the object is long and flat. Note that other cutoff values may be used, such as 0.5.
Each object is classified into the appropriate class and the distribution of accumulated pieces in the different classes is calculated based on the number of pieces and the sum of the accumulated volume of the pieces.
The obtained determination results can be directly correlated with the shape index and the flake index defined by European standards EN933: 4 and EN933: 3, respectively. That is, the data obtained provides information equivalent to the caliper and bar sieve measurements described by these standards.

(Iii) Angularity classification Angularity ("corner volume percent") is calculated for each piece based on the morphological analysis of the 3D image acquired as described above. Each piece is a collection of six classes: (1) very round, (2) round, (3) close to a circle, (4) a little corner, (5) a corner, (6) very It is classified into an appropriate class from the corner.
The classification is linearly correlated to the calculated “corner volume percent”. The distribution of accumulated pieces in different classes is calculated based on the number of pieces and the sum of the accumulated volume of the pieces.

FIG. 1 schematically shows the overall setup of a laser-camera triangulation system for optical detection and surface analysis. Two cameras 1 and 2 are shown, the laser 3 directing the laser beam towards the object 4. An arrow 5 indicates the direction of movement of the conveyor 11. Reference numeral 6 denotes an IR / VIS light source included as an option, and reference numeral 7 denotes an optional spectrophotometer for measuring the IR / VIS absorbance of the sample object 4. FIG. 2 is an overview of the preferred apparatus of the present invention. FIG. 3 is a schematic diagram of a program structure of a portion of the control system and a program for optical detection surface analysis described in more detail in Example 1. FIG. 4 shows a range intensity image without occlusion and perspective effects, representing a piece analyzed by the mathematical morphology algorithm described herein. FIG. 5 shows how the volume loss (“lost volume” or “corner volume”) of a cornered surface is determined by a “rolling ball” mathematical morphology technique. The original surface is shown in FIG. 5a, then a transformation is applied (FIG. 5b) and points that cannot be accessed by the elliptical structuring element are deleted (FIG. 5c). FIG. 6 is a cross-sectional view of three pieces derived from the capture distance image. The outer line represents the original boundary and the inner line represents the boundary after morphological opening. FIG. 6a represents a sufficiently round piece. FIG. 6b represents a slightly round piece. FIG. 6c represents a cornered piece. FIG. 6 is a cross-sectional view of three pieces derived from the capture distance image. The outer line represents the original boundary and the inner line represents the boundary after morphological opening. FIG. 6a represents a sufficiently round piece. FIG. 6b represents a slightly round piece. FIG. 6c represents a cornered piece. FIG. 6 is a cross-sectional view of three pieces derived from the capture distance image. The outer line represents the original boundary and the inner line represents the boundary after morphological opening. FIG. 6a represents a sufficiently round piece. FIG. 6b represents a slightly round piece. FIG. 6c represents a cornered piece. FIG. 7a is a scatter plot of volume loss determined according to the present invention (see Example 2) for the angularity visually assessed for each piece, sorted according to the Powers scale, and high angular It shows the variability of volume loss in the utility class. 7b is a plot of the standard deviation of the percent volume loss under each class against the manually evaluated Powers number. FIG. 7a is a scatter plot of volume loss determined according to the present invention (see Example 2) for the angularity visually assessed for each piece, sorted according to the Powers scale, and high angular It shows the variability of volume loss in the utility class. 7b is a plot of the standard deviation of the percent volume loss under each class against the manually evaluated Powers number. FIG. 8 shows the relationship between the average volume loss percentage automatically determined by the method of the present invention and the Powers number for each category. FIG. 9 shows a screen view generated by computer software for implementing the present invention. 9a shows the measurement data for one object including axis length, angularity class, shape class form, corner volume, and others. 9b shows the accumulated piece size distribution for a sample based on sample volume (left) and sample object number (right). 9c shows the distribution of sample objects in angularity and shape classes based on volume (left) and number (right). FIG. 9 shows a screen view generated by computer software for implementing the present invention. 9a shows the measurement data for one object including axis length, angularity class, shape class form, corner volume, and others. 9b shows the accumulated piece size distribution for a sample based on sample volume (left) and sample object number (right). 9c shows the distribution of sample objects in angularity and shape classes based on volume (left) and number (right). FIG. 9 shows a screen view generated by computer software for implementing the present invention. 9a shows the measurement data for one object including axis length, angularity class, shape class form, corner volume, and others. 9b shows the accumulated piece size distribution for a sample based on sample volume (left) and sample object number (right). 9c shows the distribution of sample objects in angularity and shape classes based on volume (left) and number (right).

Claims (37)

An apparatus for automatically analyzing the size and shape of a plurality of sample objects selected from minerals, rocks, gravel, natural, manufactured or recycled aggregates, etc.
a) a feeder unit for supplying said objects to the conveyor belt separated from each other in the flow;
b) an illumination source that projects a collimated beam of light across the conveyor belt;
c) image capture means comprising at least one image detector for capturing an image of the reflection of the planar beam that illuminates the object;
A control system comprising processing means for storing a computer program for controlling the mechanical part and the hardware part of the device, and a memory for storing the captured image, the control system comprising: Adapted to process a series of images of the object to automatically determine the size and shape of the object and to return parameters representing the size and shape of the object; The shape parameters include parameters including shape and angularity.
apparatus.   2. The apparatus according to claim 1, wherein the image capturing means is arranged to capture images of the reflection of the planar beam directed at the object from different angles with respect to the planar beam. An apparatus comprising two image detectors.   The apparatus of claim 1, wherein the feeder is selected from a vibrating spiral elevator or dosing feeder.   3. The apparatus of claim 2, wherein the at least two image detectors are arranged one on each side of the planar beam, each image plane being perpendicular to the direction of the conveyor belt, An apparatus positioned to have a horizontal axis parallel to the planar beam.   The apparatus of claim 1, wherein at least about 100 sample objects with a maximum width diameter of less than 100 mm per hour, preferably at least about 400 of the samples with a maximum width diameter of less than 100 mm per hour. An apparatus configured to automatically analyze an object.   The apparatus of claim 1, wherein the control system is adapted to process the series of images and the resulting three-dimensional surface image of the object based on a computer program, and Parameters, namely, the size of the object, the length of the major axis, the length of the intermediate axis, the length of the minor axis, the stretch ratio and the flake ratio, the shape class, the equivalent shape index, the equivalent flake index, the sphericity, An apparatus adapted to calculate one or more of roundness or angularity values, as well as one or more statistical distributions of said parameters for a plurality of analyzed pieces.   7. The apparatus of claim 6, wherein the control system is adapted to determine a size parameter and size distribution, shape class and shape class distribution, and angular class and angular distribution for the object. ,apparatus.   8. An apparatus according to any preceding claim, wherein the shape parameter further comprises a surface texture parameter indicative of smoothness / roughness.   9. An apparatus according to any preceding claim, wherein the control system uses a morphological algorithm to calculate a pseudo volume loss value correlated to a roundness or angularity parameter and / or class. An apparatus adapted to perform and process the series of images.   10. The apparatus of claim 9, wherein the morphological algorithm is based on the use of a structural ellipse element, wherein the element is defined for each piece to be analyzed and is substantially the same length and width as the piece. Device with a ratio.   11. Apparatus according to claim 10, wherein the proportionality constant defining the ratio between the size of the piece and the size of the structural element is in the range of about 1: 3 to 1:10, preferably about 1. A device in the range of 4 to about 1:10.   The apparatus of claim 1, wherein a light source that provides a beam of visible and / or infrared light and a reflection of the beam from an illuminated spot of the object are detected and visible and / or infrared. And a spectrophotometric detector for measuring the absorbance or reflection spectrum of the light source, wherein the control system is adapted to process the spectrum, and mineral and rock fragments and similar object types and / or An apparatus adapted to compare the spectrum with a reference spectrum and / or a spectrum value based on a reference material to classify the object according to a predetermined class system of variants.   13. The apparatus of claim 12, wherein the visible and infrared light sources provide visible and infrared light including a wavelength range of about 340 nm to about 1200 nm, the range being detected by the detector. Equipment.   14. The apparatus according to claim 1, further comprising weighting means for weighting each object. A method for determining the size and shape of an object in a plurality of objects in a sample selected from minerals, rocks, gravel, natural, manufactured or recycled aggregates, etc. Determining the shape includes at least determining a shape parameter or class that indicates whether the piece is long and / or flat, and determining an angularity, the method comprising:
Feeding the objects spaced apart from each other onto a conveyor belt;
Illuminating sequentially flowing objects with a collimated beam of light across the direction of the conveyor;
For each object, at least one image detector is used to illuminate the object so that a series of images are acquired and stored at a regular minimum frequency based on a predetermined speed of the conveyor Capturing an image of the diffuse reflection of the planar beam and storing the image in memory;
Processing the series of images for the object for each object passing under the planar beam of light to obtain three-dimensional surface data or contour maps and topographical data representing the object;
Determining a size parameter and a shape parameter for the object based on the obtained data, wherein the shape parameter indicates whether the piece is long and / or flat, and A method comprising / and classification and angularity parameters and / or classification.   16. The method of claim 15, comprising capturing an image using at least two image detectors, one on each side of the collimated beam of light.   16. The method of claim 15, wherein the size and shape data including shape data indicating longness and flatness and angularity / roundness data for each object in the plurality of objects is edited. And calculating parameters representing mean and variation values of the size and shape of the analyzed object in the sample.   18. A method according to claim 17, wherein the size of the object is represented by an ellipse calculated from the minor and intermediate axes of the object.   18. A method according to claim 17, wherein the size of the object is indicated by a calculated volume based on the measured dimensions and a predetermined shape.   20. A method as claimed in claim 19, wherein an ellipse is calculated from three axes representing the external dimensions of the object, the ellipse generating a size distribution for a plurality of objects to be analyzed. A method that provides an approximation of the volume of the.   21. A method according to any of claims 15 to 20, wherein the shape parameter and / or classification is such that the object is substantially spherical, substantially flat, or substantially long. A method comprising shape classification into at least four classes that indicate whether they are or are substantially long and flat.   A method according to any of claims 15 to 21, wherein an elliptical structuring element having an axis proportional to the axis of the object is used in a mathematical morphology algorithm to determine the angularity of the object. The way it is.   23. The method of claim 22, wherein a proportionality constant defining a ratio between the size of the piece and the size of the structural element is in the range of about 1: 3 to 1:10, preferably about 1. A method in the range of 4 to about 1:10.   24. The method of claim 23, wherein the proportionality constant is in the range of about 1: 4 to about 1: 8.   25. A method according to any of claims 15 to 24, wherein the angular parameter and / or classification comprises a classification scheme having a plurality of classes.   26. A method as claimed in any of claims 15 to 25, illuminating each object supplied to the conveyor with a visible and / or infrared radiation beam, and the beam from an illuminated spot of the object. Measuring the visible and / or infrared absorbance or reflection spectrum, and measuring the spectrum, a reference spectral value to determine petrological information or other composition information about the object And comparing with the method.   27. The method of claim 26, wherein the petrological information includes classification of the object according to a predetermined class system of petrological rock types and / or variants. Loaded into a computer to control and analyze the acquisition of image data for multiple objects selected from minerals, rocks, gravel, natural, manufactured or recycled aggregates, etc. A possible computer program product comprising program instruction means for instructing a computer processor when loaded into a computer and executed, the program instruction means comprising:
Input data from at least one image detector configured to capture a sequential image of a reflection of a planar coherent beam that illuminates across the direction of flow of the object moving at a predetermined speed with a minimum predetermined frequency. To receive
Storing the image for each object and processing the image to obtain data indicating the size of the object and three-dimensional surface data or contour map and topographical data representing the object;
Instructing to determine one or more size and shape parameters for the object based on the obtained data, wherein the shape parameters are long and / or flat pieces Including shape parameters and / or classification to indicate whether or not, angular / roundness parameters and / or classification,
Computer program product.   29. The computer program product according to claim 28, wherein size and shape data including long and flat shape data and angularity / roundness data for each object in the plurality of objects. A computer program product further comprising program instruction means for calculating parameters representing mean and variation values of the size and shape of the plurality of objects.   30. The computer program product of claim 28 or 29, wherein the shape parameter and / or classification is such that the object is substantially spherical, substantially flat, or substantially long. A computer program product comprising shape classification into at least four classes that indicate whether they are or are substantially long and flat.   31. A computer program product according to any of claims 28 to 30, wherein the size of the object is indicated by a volume calculated based on measured dimensions and a predetermined shape.   32. The computer program product of claim 31, wherein an ellipse is calculated from three axes representing external dimensions of the object, and the ellipse generates a size distribution for a plurality of objects to be analyzed. A computer program product that provides an approximation of the volume of the object.   33. A computer program product according to any of claims 28 to 32, wherein an elliptical structuring element having an axis proportional to the axis of the object to be analyzed is mathematical to determine the angularity of the object. A computer program product used in a genetic morphology algorithm.   34. The computer program product of claim 33, wherein a proportionality constant defining a ratio between the size of the piece and the size of the structural element is in the range of about 1: 3 to 1:10, preferably Is a computer program product in the range of about 1: 4 to about 1:10.   35. The computer program product of claim 34, wherein the proportionality constant is in the range of about 1: 4 to about 1: 8. 36. A computer program product according to any of claims 28 to 35, comprising:
Receiving spectrophotometric input data from a spectrometer configured to detect reflections of visible and / or infrared beams that illuminate the object;
A program product further adapted to compare the data with reference spectral values to determine petrological or other compositional information about the object.   37. The computer program product of claim 36, wherein the petrological information includes classification of the object according to a predetermined class system of petrological rock types and / or variants. .

Images